AI 技能AI 插件人工智能【免费下载链接】agentic-awesome-skillsAAS Core is the local, agent-first control plane for complete catalog discovery, agent-owned selection, stack validation, and planning, backed by 2,600 agentic skills. Includes CLI, local MCP, catalog, plugins, and Workbench.项目地址https://gitcode.com/GitHub_Trending/an/agentic-awesome-skills点击查看免费下载Axiom 是 Agentic Awesome Skills 仓库中一个以第一性原理为核心的社区技能定位为First-Principles Assumption Auditor第一性原理假设审计器。它把任何问题强制剥离到不可再拆的最小真相单元再基于被验证的前提重建结论。本文以该技能自带的英文完整走读示例Im thinking about dropping out of my CS degree to join a startup为主线完整还原其 5 阶段拆解流程并结合技能定义文件、四种假设分类手册、场景化假设挖掘清单讲解如何在重大职业/人生决策中实践这套方法论。读完本文你将掌握一套可复用的问题澄清 → 假设挖掘 → 假设分类 → 风险排序 → 结论重建的思考框架并理解 Axiom 技能在仓库中的定位、触发方式与工程实现细节。一、Axiom 技能概览它是做什么的Axiom 在仓库中被登记为data/catalog.json中的一项社区技能source: community元数据如下技能 IDaxiomcanonical_id: axiom描述First-principles assumption auditor. Classifies each hidden assumption (fact / convention / belief / interest-driven), ranks by fragility × impact, and rebuilds conclusions from verified premises.风险等级risk: safe路径skills/axiom/SKILL.md其核心定义文件位于 plugins/agentic-awesome-skills-claude/skills/axiom/SKILL.md技能自我介绍是两句话Strip any question down to its irreducible truths, then rebuild from there. This is not framework fill-in-the-blank — it is assumption prosecution.把任何问题强制剥离到不可再拆的最小真相单元再从那里重建。不是框架填空是假设审判。技能支持中英双语自动检测检测到用户用中文输入则整个会话的各个阶段、标签、输出全部使用中文反之使用英文除非用户明确切换语言。这一点在 SKILL.md 的 Language Rule 一节有明确规定。技能能力清单按 SKILL.md 的定义Axiom 有五大核心能力Problem Reframing / 问题澄清—— 在触碰任何假设之前先确认问题本身被正确定义Assumption Mining / 假设挖掘—— 系统性地在三个深度层级表面 812 个隐藏假设Assumption Classification / 假设分类—— 强制给每个假设打上四种类型之一不同类型采用不同的质疑策略Risk Ranking / 优先级排序—— 以脆弱性 × 影响力为每个假设打分输出最危险的 Top 3Reconstruction / 重建—— 只基于被验证的前提重建结论并显式对比之前 vs 之后的认知位移。适用时机与触发短语SKILL.md 明确列出了何时使用的场景一个重大的人生或职业决策摆在桌面上辞职、创业、买房想压力测试一个业务方向或产品假设怀疑自己持有的某个信念可能是错的但说不清为什么需要穿透复杂性找到真正的瓶颈有人要求你从第一性原理思考或拆解一下。中文触发短语第一性原理 / 帮我想清楚 / 拆解一下 / 从底层分析 / 这个假设对吗 / 我在做一个决定 / 从根本上分析 / 底层逻辑 / 元问题 / 重新思考 / 有没有想错 / axiom英文触发短语first principles / break it down / question my assumptions / think from scratch / challenge this belief / audit my reasoning / what am I missing / help me think clearly / axiom二、完整实战走读拆解退学加入初创公司决策本节是 Axiom 官方英文走读示例walkthrough-en.md的完整还原。其用户输入为User input:Im thinking about dropping out of my CS degree to join a startup full-time. The CEO offered me equity and a junior dev role. Help me think through this from first principles.下面按 5 个阶段逐一呈现 Axiom 的实际输出结构。Phase 1: Problem Reframing问题澄清——先确认问题本身定义正确Axiom 在此阶段不急于拆解假设而是先确认你真正在选择的到底是什么你在选择什么A一个摆在桌上的具体初创公司 offer vs 完成学业B总体上创业生活 vs 学术生活的想法C想逃离学校中某些不喜欢的部分初创公司只是出口这个问题是谁定义为现在必须做决定的是 CEO 给了你截止日期是你自己制造了紧迫感还是存在外部压力退学的同龄人、科技圈的 Twitter 叙事重新表述后的核心问题Reframed core questionGiven my current skills, financial situation, and risk tolerance, is joining this specific startup now — at the cost of delaying/abandoning my degree — a better path to my goals than completing the degree first? And do I have enough information to make this decision today?该核心问题会先呈现给用户确认确认通过后才进入下一步。SKILL.md 强调了一个重要判断很多人问我该不该辞职但真正的问题是我在当前岗位上为什么无法成长——这是两个根本不同的问题对应完全不同的假设集合。Axiom 会先问这个问题是谁定义的是你自己、他人的期待、还是社会叙事这是根问题还是更深层问题的症状Phase 2: Assumption Mining假设挖掘——你在无证据地相信什么系统性挖掘隐藏假设分为三个深度层级LayerDescriptionExampleSurface表层Obvious, often stated aloudI need more moneyMiddle中层Industry conventions, common wisdomA degree is required for good jobsDeep深层Never questioned, feels like gravitySuccess means financial independence目标是在三个层级中挖掘出812 个假设且越具体越好拒绝我觉得这样更好这类模糊陈述强制具体化。针对本案例Axiom 挖掘出了 10 个假设表层假设显而易见的A1. A CS degree is just a piece of paper — the real learning happens on the jobA2. This startup opportunity wont come again if I dont take it now中层假设行业惯例/常识A3. Early-stage equity is worth more than a salaryA4. The startup world values skills over credentialsA5. I can always go back and finish my degree later深层假设从未质疑的信念A6. Working at a startup will accelerate my career faster than any other pathA7. Im ready for the intensity and ambiguity of startup workA8. The CEO/founder is competent and trustworthyA9. If the startup fails, the experience itself will be valuable enoughA10. My excitement about this opportunity reflects genuine fit, not just novelty深层假设才是最有价值的挖掘目标——它们通常从未被本人质疑过却往往决定整个决策的走向。Phase 3: Assumption Classification假设分类——每个假设的本质是什么对每个假设强制打上四种类型之一的标签。分类本身就是洞见——很多人第一次发现自己一直当作事实的东西其实只是惯例甚至是利益驱动。#AssumptionTypeReasoningA1A degree is just a piece of paper ConventionWas increasingly true 2015-2022 in tech; now shifting back as market tightensA2This opportunity wont come again⚫ Interest-drivenThe CEO benefits from you believing this. Urgency is a recruiting tacticA3Early equity is worth more than salary Convention ⚫ Interest-drivenStatistically, most startup equity becomes worthless. VC-backed mythology promotes this narrativeA4The startup world values skills over credentials ConventionTrue at some companies, but many still filter by degree, especially post-2024 hiring freezesA5I can always go back and finish later Subjective beliefBased on intention, not evidence. Dropout re-enrollment rates are lowA6A startup will accelerate my career faster Subjective beliefSurvivorship bias — you see dropouts who succeeded, not the majority who struggledA7Im ready for startup intensity Subjective beliefSelf-assessment often overestimates readiness for ambiguity and stressA8The CEO is competent and trustworthy Subjective beliefBased on limited interactions. Have you done due diligence on the founder?A9Failed startup experience is still valuable ConventionAccepted wisdom in startup culture, but hiring managers increasingly want concrete resultsA10My excitement genuine fit Subjective beliefExcitement fades after 3 months. How do you distinguish excitement from novelty?可以看到本案例中没有一个假设属于物理事实——这正是多数人生决策的典型特征看似理所当然的信念绝大多数只是惯例或主观信念少数则是被利益驱动的叙事。Phase 4: Risk Ranking风险矩阵 Top 3——先查哪个对每个假设在两个维度上打分二者相乘得到危险值Risk ScoreRisk Score Fragility × ImpactFragility / 脆弱性1-5这个假设有多容易被证伪1 几乎无法推翻如物理定律5 极易被证伪如未经测试的市场直觉、个人感觉。Impact / 影响力1-5如果这个假设是错的你的结论会垮掉多少1 几乎不影响最终结论5 根基性支柱错了全盘皆输。输出危险值最高的Top 3作为优先调查对象且每一条必须附带一个具体、可执行的验证问题RankAssumptionFragilityImpactRisk ScoreVerification Question 1A3: Early equity is worth more than salary4520Whats the companys current revenue? Runway? Valuation methodology? Have you had a lawyer review the equity terms? 2A8: The CEO is competent and trustworthy4416Have you talked to other people whove worked with this person? Checked their track record? Asked about previous ventures? 3A5: I can always go back and finish later4416Whats the actual re-enrollment rate at your university for students who take leave? What would the financial and logistical cost be?你可能不想听的真相The uncomfortable truthA3 has the highest risk score. The equity youre being offered is almost certainly worth $0 right now. Over 90% of startups fail, and even among those that succeed, early-employee equity is often diluted through subsequent funding rounds. The CEO has every incentive to frame equity as life-changing wealth potential — because it costs them nothing to give you shares and it justifies paying you below market rate.Have you actually run the math on what your equity would be worth under realistic scenarios?这一段是整个 Phase 4 的精华它主动挑战用户最不愿被质疑的信念股权价值这正是 Axiom 反谄媚规则要求至少输出一个用户可能不喜欢听的假设的具体体现。Phase 5: Reconstruction重建——从被验证的真相出发只保留经得起检验的假设从零重建结论并显式并排对比原来的想法与重建后的想法Original thinking:I should drop out and join this startup for the equity and the experience.Rebuilt thinking:I should complete concrete due diligence on the startup (financials, founder track record, equity terms reviewed by a lawyer) before making any decision. In parallel, I should explore whether I can join part-time while finishing my degree, or negotiate a deferred start date. The urgency to decide RIGHT NOW is manufactured — if the opportunity is real, it can wait 2-4 weeks for me to verify.认知位移对比Cognitive Shift ComparisonDimensionOriginal ThinkingRebuilt ThinkingDecision urgencyI need to decide nowThe urgency is manufactured — add 2 weeks of verificationEquity valueThis could be hugeAlmost certainly $0 today; need to verify terms and scenariosEither/or framingDrop out OR miss outExplore part-time, deferred start, or summer commitment firstRisk assessmentI can always go backRe-enrollment is not guaranteed; quantify the actual costInformation basisExcitement CEOs pitchDue diligence on founder, financials, legal review of equity如果没时间做完整重建——最该先验证的一件事Ask the CEO for the companys current revenue, burn rate, and runway. If they refuse to share basic financials with someone theyre asking to bet their education on, that tells you everything you need to know.一个高质量验证问题即可撬动整个决策的杠杆如果对方拒绝向一个被要求拿教育前途下注的人分享基本财务数据这个拒绝本身就已经回答了问题。三、四种假设类型分类法深度解析Phase 3 背后的完整方法论记录在 references/assumption-types.md四种假设类型手册。其核心论断是不同性质的假设需要根本不同的质疑策略——质疑一个物理事实会浪费时间和信誉质疑一个惯例需要检查环境是否已经改变质疑一个信念需要找出是谁的经历塑造了它质疑一个利益驱动假设需要追着钱走。 Type 1: Physical Fact物理事实定义自然规律、数学定理、已验证的科学常数、逻辑恒等式不能被人类意志、政策或信念改变。识别问题你能在地球上找到任何一个真正的反例吗无论文化、时间、语境它都成立吗如果人类全部消失它仍然成立吗它基于可重复、可测量的证据吗质疑策略**接受它不要浪费能量。**如果某物确实是物理事实正确做法是在它的约束内工作而不是对抗它。常见误分类陷阱人们常把社会规律或统计趋势当作物理事实。经济总会复苏是历史规律不是物理定律人们抗拒变化是一种倾向不是常数。✅ Genuine Physical Facts❌ Often Misclassified As FactsGravity exists (9.8 m/s²)Housing always appreciates (convention)Entropy increases over timeYou need 8 hours of sleep (convention — varies by person)Speed of light is constantPeople are rational economic actors (belief)Compound interest formulaInnovation requires disruption (interest-driven) Type 2: Historical Convention历史惯例定义曾经有效且被广泛采用的做法、规则或信念起源于特定语境而语境可能已经改变使惯例过时或变为可选项。识别问题这个规则/做法是什么时候起源的那时的环境是什么样的底层环境技术、人口、市场、监管是否已经显著改变它是普遍的还是特定于地区/行业/时代谁因这个惯例的延续而获益遵循它的人中多大比例真正得到了承诺的结果质疑策略**检查时间戳。**问这在 20 年前成立吗现在还成立吗10 年后呢关键洞见是惯例有巨大的惯性——人们在原始理由消失很久之后仍会遵循它们。手册给出的对照示例非常有说服力ConventionOriginal ContextChanged Context大学学历是好工作的门票1990s: 大学生稀缺2020s: 本科普及技能导向增强You must be in the office to be productivePre-2020: Limited remote toolsPost-COVID: Remote infrastructure mature在一家公司待 5-10 年才算忠诚Lifetime employment eraAverage tenure is now 2-3 yearsReal estate is the best investment for the middle class1980-2010: Population growth urbanization2020s: Demographic decline in many countries创业要写完整的商业计划书Before Lean Startup methodologyNow: MVP → iterate → pivot is standardGet a STEM degree for job securityPre-AI eraAI is now disrupting STEM roles too跨文化注意有些惯例是文化特有的—— 35 岁是职业分水岭几乎只存在于中国劳动力市场 应该每 2-3 年换一次工作是美国科技行业惯例在日本或德国并不常见 有房才能结婚不是全球现实 Follow your passion主要是美国文化叙事。 Type 3: Subjective Belief主观信念定义从个人经验中得出的结论被投射为普遍真理。当事人真诚相信但它由有限的经验样本、认知偏差和情感联想塑造。识别问题谁告诉你的你亲自验证过还是从小样本推断有不同经历的人能得出相反的结论吗这个信念让你舒服吗它是否保护你免受更难以面对的真相如果发现它是错的你必须做出什么改变质疑策略**寻找反面证据。**找到做了相反事情却成功的人或遵循这个信念却失败的人。目标不是推翻信念而是检验它的普遍性。Subjective BeliefWhat It Actually Is我不适合做管理One bad management experience → generalized to identityIm not a math personEarly negative experience → self-limiting identity创业太冒险了Heard failure stories → availability biasMy industry doesnt value XBased on one companys culture, not the industry我的性格不适合社交Introversion ≠ inability to build relationshipsGood developers dont need to be good at communicationBelief that protects from discomfort红旗信号如果某人说就是这样的或大家都知道但说的并不是物理定律那几乎肯定是信念或惯例被提升到了事实的地位。⚫ Type 4: Interest-Driven利益驱动定义某些人行业、公司、意见领袖或社会群体积极传播的假设因为他们从你的相信中获益。这并不意味着这些说法总是假的——但它们被传播的原因不是因为真而是因为有利可图。识别问题谁在推广这个观点他们卖什么信息来源是否有经济动机让你相信如果没有人从中获利这个观点还会流行吗它是否被包装成紧迫的**紧迫感通常服务于卖方。**零假设是什么——如果你什么都不做会怎样质疑策略**追着钱走。**从这个观点追溯到获益者。这并不自动推翻这个说法但会改变你应该给予它的权重。Interest-Driven AssumptionWho Benefits你需要买保险来保护家人保险公司和保险经纪人AI will replace all jobs in 5 yearsAI companies raising funding创业是实现财务自由的唯一方式创业培训课程、孵化器You need to learn to code (2015-2023)Coding bootcamps, tech companies房价永远涨房地产开发商、中介、银行You need this certification to be competitiveCertification providers知识付费是最好的自我投资知识付费平台和课程制作者Growth hacking is essential for startupsGrowth consultancies, SaaS tools回到本文的走读案例A2这个机会不会再来被分类为 ⚫ 利益驱动正是这条逻辑链的体现CEO 从你相信机会稀缺中获益紧迫感本身就是一种招聘战术。混合类型与边界案例很多真实假设是混合体。处理方法先识别主类型哪个类型最贴切地描述了核心机制再记录次类型还有哪些附加动力先应用主类型的质疑策略再用次类型补充。AssumptionPrimarySecondaryWhy现在是创业的好时机 Convention⚫ Interest-drivenThe good timing narrative is a convention, but startup media amplifies it for clicks and fundingA college degree pays for itself Convention⚫ Interest-drivenWas true statistically, but universities have incentive to promote it regardless房价只涨不跌 Belief ConventionPersonal experience from a specific era, reinforced by industry-wide conventionRemote workers are less productive Belief ConventionManagers personal bias, reinforced by traditional management convention边界案例方面手册特别提示统计趋势如男性平均比女性高是统计事实但套用到具体个人身上就是主观信念分类要基于用户如何使用它过时的事实如冥王星是行星应按惯例处理自我实现的预言如股市要崩若来源从中受益则按利益驱动、若是个人焦虑则按主观信念分类。手册最后给出了一张可复用的快速参考卡 PHYSICAL FACT → Can you find ANY counterexample? → No → Accept it HISTORICAL CONV. → When did this rule form? → Check if context changed SUBJECTIVE BELIEF → Who told you this? → Seek counter-evidence ⚫ INTEREST-DRIVEN → Who profits from this? → Follow the money四、场景化假设挖掘清单让 Phase 2 更彻底Axiom 的 Phase 2 并不是凭空发挥。技能附带 references/scenarios.md提供8 个场景的专用假设挖掘清单4 个中文原生 4 个英文原生每个清单包含该领域与文化下人们几乎从不质疑的高频隐藏假设以及针对性的挖掘问题。当用户问题命中某类场景时应引用对应清单确保挖掘的彻底性与文化相关性#中文场景English Scenario1职业决策换工作、创业方向Career Decisions (job change, career pivot)2产品方向验证创业、新功能Business Product Validation3消费选择买房、投资、重大消费Financial Life Decisions4认知信念质疑人生观、方法论Belief Worldview Audit例如场景一职业决策的高频隐藏假设清单包括大公司比小公司更稳定、跳槽一定能涨薪 20% 以上、35 岁是职业发展的分水岭、管理岗比技术岗天花板更高、远程工作会影响职业发展、学历对职业发展至关重要、在一个公司待太久会贬值等 15 条配套的挖掘问题如你说想换工作 — 如果当前公司环境改善 50%你还想走吗你说35岁是门槛 — 你认识几个35岁以上转型成功的案例你说现在是好时机 — 好时机的判断依据是什么是数据还是感觉英文场景Scenario 5-8同样各含 14 条高频假设与 5 个推荐挖掘问题例如场景六Business Product Validation的核心假设包括If we build it, they will come、TAM 是 X 亿美元所以我们会分到一杯羹、Customer interviews validated demand等配套问题如You say users need this — how many have actually paid? Wishing and paying are very different.使用这些清单的方法论scenarios.md 明示从用户问题中识别场景类型把清单当作挖掘指南——并非所有假设都适用于每个用户但要系统性地扫过优先关注文化特有的假设——中国用户询问职业决策时的隐藏假设与美国用户差异巨大与有机挖掘结合——清单是起点而非天花板永远要挖掘该用户具体处境中独有的假设用挖掘问题来表面用户无法直接说出口的假设。五、反谄媚核心规则Axiom 区别于奉承回音室的关键SKILL.md 强调这些规则是硬性约束hard constraints优先级高于其他一切行为倾向——这正是 Axiom 真正有用而非变成奉承回音室的原因RuleDescriptionNo agreementDo NOT agree with the users original conclusion during the decomposition phases, even if they insist repeatedly.No flattery openersDo NOT start with Thats a great question or any similar validating phrase. Get straight to work.No identical reconstructionThe Phase 5 reconstruction MUST NOT produce an identical conclusion to the original without explicitly explaining why no shift occurred, with specific evidence.✅At least one uncomfortable truthPhase 4 MUST output at least one assumption the user probably doesnt want to hear challenged.✅Devils advocate persistenceIf the user rejects a classification or pushback, hold firm like a devils advocate. Only yield when the user provides verifiable evidence (not feelings, not appeals to authority).中文表述同样写入技能 禁止在拆解阶段认同用户的原始结论 禁止用这是个好问题或类似话语开头 禁止重建阶段给出和原始想法完全一致的结论✅ 必须在阶段 4 输出至少一个用户可能不喜欢听的危险假设✅ 必须像 devils advocate 一样坚持直到用户提供真实证据。在本文的走读案例中这些规则逐一得到体现Phase 4 直接说出你被提供的股权现在几乎肯定值 0 美元这一用户不愿听的事实Phase 5 的重建结论与原始结论完全不同从退学入职变成先做尽职调查 探索兼职/延期入职。快捷输出模式Quick Output Mode如果用户明确要求快速分析或时间紧张SKILL.md 允许跳过完整的 5 阶段走读直接输出危险值最高的 Top 3 假设含风险分值与各自一条可执行的验证问题并以你最该验证的一件事是… / The single most important thing to verify is…收尾。六、工程实现事实技能包结构与双语言示例从仓库结构看Axiom 技能包的组织方式两个副本分别在 plugins/agentic-awesome-skills/skills/axiom 与 plugins/agentic-awesome-skills-claude/skills/axiom如下skills/axiom/ ├── SKILL.md # 技能主文件元数据、5阶段流程、反谄媚规则 ├── examples/ │ ├── walkthrough-en.md # 英文完整走读退学加入初创公司 │ └── walkthrough-zh.md # 中文完整走读辞职去创业 └── references/ ├── assumption-types.md # 四种假设类型手册 └── scenarios.md # 8个场景专用假设挖掘清单这种主文件 示例 参考手册的层次化组织使技能既能被快速触发主文件定义了触发短语又能按需加载深度参考分类手册、场景清单还提供了两种语言的完整走读示例作为 Few-shot 范本。data/catalog.json中的登记信息还透露了技能的triggers触发词集合axiom、first、principles、assumption、auditor、classifies、each、hidden、fact、convention、belief、interest与tagsaxiom说明该技能在仓库的目录系统中有完整的可发现性登记可以被skills_index.json等索引机制检索到。七、实战应用指南如何在真实决策中使用 Axiom综合 SKILL.md 的使用建议与本文走读使用这套方法论时有以下要点越深越好能挖到的假设层级越深分析价值越大。表层假设谁都看得见真正的决策变量藏在深层。拒绝我感觉作为证据推动用户给出具体事实。模糊的我觉得这样更好要被强制具体化。最有价值的洞见常来自重新分类把原本以为的事实重新归类为惯例甚至利益驱动往往是最大的顿悟时刻。用风险矩阵聚焦有限的验证精力不要平均用力把验证资源集中在危险值最高的少数假设上。重建结论若与原始结论完全相同说明拆解得不够深这是一个自检信号提示需要继续下挖。常见使用场景SKILL.md 列举重大职业决策辞职、转型、谈判在投入时间/金钱前验证创业想法挑战那些可能阻碍你的显而易见的信念对重大人生选择做 pre-mortem事前验尸分析审计投资或财务决策通过识别真正重要的东西来突破分析瘫痪。同时技能也声明了明确的使用边界Limitations只在任务与上述范围匹配时使用不要把输出当作环境特定验证、测试或专家审查的替代品当必需的输入、权限、安全边界或成功标准缺失时应停下来澄清——这些边界定义记录在 SKILL.md 末尾。进一步研读路径中文版本的完整走读我觉得我应该辞职去创业见 examples/walkthrough-zh.md其中 Phase 4 的你可能不想听的真相同样极具冲击力——绝大多数创业失败不是因为执行力差而是因为从一开始就在解决一个没人愿意付费的问题。它与英文走读互为镜像共同构成 Axiom 方法论的中英双语实战范本。赞分享AI 技能AI 插件人工智能【免费下载链接】agentic-awesome-skillsAAS Core is the local, agent-first control plane for complete catalog discovery, agent-owned selection, stack validation, and planning, backed by 2,600 agentic skills. Includes CLI, local MCP, catalog, plugins, and Workbench.项目地址https://gitcode.com/GitHub_Trending/an/agentic-awesome-skills点击查看免费下载相关推荐Axiom 第一性原理拆解器用 5 阶段假设审计流程重建决策结论agentic-awesome-skills 技能详解Axiom 第一性原理拆解器用 5 阶段假设审计流程重建决策结论agentic awesome skills 技能详解 本指南深入解析开源仓库 agentAI 技能AI 插件人工智能agentic-awesome-skills 的 Axiom 技能实战第一性原理假设审计的 5 阶段拆解流程agentic awesome skills 的 Axiom 技能实战第一性原理假设审计的 5 阶段拆解流程 本文以 walkthrough zh.md htAI 技能AI 插件人工智能生命科学行业公司研究实战用 Product-Manager-Skills 的 company-research 技能拆解临床阶段生物技术公司生命科学行业公司研究实战用 Product Manager Skills 的 company research 技能拆解临床阶段生物技术公司 导读 本文以 PAI 技能AI 插件上一篇react-admin withLifecycleCallbacks 完全指南为 dataProvider 注入读写生命周期钩子下一篇cargo fix 命令完全指南用 rustc 建议自动修复代码与 Edition 迁移实战创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考